Paper template for HIC2004
th
Extended Abstract
11 ISE 2016, Melbourne, Australia
DRIVERS OF THE DISTRIBUTION OF A DOMINANT RIPARIAN TREE SPECIES (EUCALYPTUS COOLABAH) ON A DRYLAND RIVER SYSTEM, DIAMANTINA RIVER, AUSTRALIA JUSTIN F. COSTELLOE, JANE LEEDER, MARCUS STRANG Department of Infrastructure Engineering, University of Melbourne Melbourne, VIC, 3010, Australia
Coolibah (Eucalyptus coolabah) is an iconic Australian tree species and the dominant riparian species in many arid zone river systems, such as in the Lake Eyre Basin. The successful germination and growth (i.e. recruitment) of these trees, which occurs in response to floods, is critical to the ecology of dryland rivers. Data on the position, elevation and size of coolabahs were collected from field transects in 2014 and related to soil data and the Geoscience Australia satellite mapping of inundation frequencies. Multivariate analysis showed that mature trees were most highly correlated with frequently inundated (i.e. near bank) positions and younger trees with less frequently inundated (i.e. distal) positions. Field observations showed that this relationship was complicated by small-scale topographical features that were favoured sites for recruitment, such as edges of „flood runners‟ (i.e. shallow floodplain channels that connected with the primary channels) and sand dune – floodplain boundaries. Coolabahs could grow in a wide range of soil types but favoured silt-sand soil types and rarely grew in floodplains dominated by cracking clays with well-developed gilgai structures. Relatively few recruits were identified from the sequence of large flood years in 2009-2011. There was evidence of considerable grazing pressure on the younger coolabahs and also plant moisture stress from the subsequent dry years of 2013-2015. The subdued recruitment response to the 2009-2011 flood sequence highlights the vulnerability of dryland riparian vegetation to grazing and climate change effects. 1
INTRODUCTION
Riparian and floodplain tree communities are often the dominant woodland assemblage in the arid zone. These communities have high ecological value in providing habitat, food sources and shade, and are a significant influence on the ecology of arid regions [1]. Coolabah (Eucalyptus coolabah) is an iconic Australian tree species and the dominant riparian species in many arid and semi-arid zone river systems, such as in the Lake Eyre Basin (LEB). The successful germination and growth (i.e. recruitment) of these trees, which occurs in response to floods [2, 3], is critical to the ecology of these dryland rivers. Understanding the drivers of recruitment success is vital to determining how these systems respond to current and future pressures driven by grazing, flow regulation and climate change. Previous work has examined regeneration characteristics of E. coolabah on the regulated rivers of the Darling River catchment, where clearing of this species has commonly occurred [3]. The unregulated rivers and the largely natural vegetation communities of the LEB provide an excellent system for analysis of natural patterns of tree distribution. This study examines field data on tree distribution, flooding characteristics and soil properties to identify drivers of successful germination and growth (i.e. recruitment) of E. coolabah in a large unregulated arid zone river, the Diamantina River of the LEB. The Diamantina River is one of the major rivers of the LEB and is the major contributor to flow into Lake Eyre North, the terminus of the endorheic LEB. It has a summer-dominant, intermittent flow regime and shows a variety of channel-floodplain morphologies, ranging from meandering single channel to vast areas of poorly channelized floodplain dominated by cracking clay sediments [4]. 2
METHODS
2.1 Field data collection Fieldwork was undertaken during a field trip in May 2014. Data on the distribution of E. coolabah trees were collected from 19 belt transects on the Diamantina floodplain. Transects typically began at the main channel and were directed approximately perpendicular to the channel direction and were of varying length (200-1200 m). All E. coolabah trees located within the 10 m width of the transect were measured for their location and elevation (by differential GPS), number of stems, circumference of the largest stem above ground level,
maximum height of tree canopy, maximum width of tree canopy and reproductive status (i.e. fruiting or not fruiting). In addition, between 2-6 soil profiles where drilled using a hand auger on each transect, totaling 54 soils profiles. Profiles were between 0.3-0.5 m deep with soil samples collected every 0.1 m. The soil profile locations were chosen to characterize soil conditions where E. coolabah trees were relatively abundant and where they were absent (i.e. no E. coolabah within 30 m radius of the soil profile). The Geoscience Australia Water Observations from Space (WOfS) dataset provided a measure of the frequency of inundation over the study area using the Landsat record. This dataset has a 25 m spatial resolution and was used to assign mean frequency of inundation values to individual tree locations in each transect (see Figure 1). Laboratory analysis was subsequently completed on the soil samples and included: gravimetric water content of a 100 g sample dried at 100°C for 24 hours; electro-conductivity (EC) and pH measurements taken from a 1:5 mix of dried soil sample (10 g) and deionized water (50 ml); plasticity measurements using a „rolltest‟ method. A small amount of the soil was mixed with a few drops of water until soil moisture content reached field texture (i.e. just below sticky point). The resulting mix was rolled into a cylinder by hand to a length of at least 40 mm. The cylinder was then held by the end to determine if it can support its own weight. The degree of plasticity categorises the soil samples, where a diameter of 2 mm gives a very plastic soil (scored 5) and a diameter of 6 mm is non-plastic (scored 1). Figure 1. Location of transects along the Diamantina River overlaying the Water Observations from Space (WOfS) mapping of frequency of inundation mapping.
Table 1. K-means cluster analysis results showing five clusters and mean characteristics for live E. coolabah individuals within transects. Cluster Circum. (m) Crown (m) Height (m) n
A
B
C
D
E
0.08 0.80 1.19 533
0.23 2.17 3.15 188
0.52 4.43 5.78 159
1.02 8.45 6.70 87
1.38 13.33 9.09 51
2.2 Data analysis Three different methods were used to analyse the datasets; principal component analysis (PCA), K-means cluster analysis and the Kolmogorov-Smirnov (KS) test. The PCA analysis was used to minimize the effects of covariance in identifying variables correlated with the tree distribution, the K-means cluster analysis identified natural groupings within the tree dataset, while the KS test analyses for differences between two populations without assumptions of normal distribution. A PCA was initially performed on each of the three dependent tree measurement variables (diameter, crown and height) and 16 independent variables (soil characteristics including salinity, pH, plasticity, soil water content; inundation frequency; elevation above local channel thalweg; distance to other trees). The variables were range-standardised to remove noise in the data and remove any scaling influences. Outliers were identified and removed from the dataset following an initial PCA analysis. The results of the covariance matrix were used to identify poorly correlated variables; which were also subsequently removed from the data in an attempt to increase the variability captured by the first three principal components. A PCA was then re-performed on the reduced dataset. Two sample, non-parametric Kolmogorov-Smirnoff (KS) tests were performed to further analyse the soil data. Two populations of soil data were tested for significance between variables; „soil samples with trees‟ and „soil samples without trees‟.
3
RESULTS AND DISCUSSION
The May 2014 field trip measured 1057 individual trees with circumferences ranging from 0.01-3.30 m, heights from 0.10-14.00 m and crown diameters from 0.02-24.30 m. Seedlings related to recent floods in the 2009-2012 period were not obvious to identify as the majority of individuals with short heights (e.g.
Extended Abstract
11 ISE 2016, Melbourne, Australia
DRIVERS OF THE DISTRIBUTION OF A DOMINANT RIPARIAN TREE SPECIES (EUCALYPTUS COOLABAH) ON A DRYLAND RIVER SYSTEM, DIAMANTINA RIVER, AUSTRALIA JUSTIN F. COSTELLOE, JANE LEEDER, MARCUS STRANG Department of Infrastructure Engineering, University of Melbourne Melbourne, VIC, 3010, Australia
Coolibah (Eucalyptus coolabah) is an iconic Australian tree species and the dominant riparian species in many arid zone river systems, such as in the Lake Eyre Basin. The successful germination and growth (i.e. recruitment) of these trees, which occurs in response to floods, is critical to the ecology of dryland rivers. Data on the position, elevation and size of coolabahs were collected from field transects in 2014 and related to soil data and the Geoscience Australia satellite mapping of inundation frequencies. Multivariate analysis showed that mature trees were most highly correlated with frequently inundated (i.e. near bank) positions and younger trees with less frequently inundated (i.e. distal) positions. Field observations showed that this relationship was complicated by small-scale topographical features that were favoured sites for recruitment, such as edges of „flood runners‟ (i.e. shallow floodplain channels that connected with the primary channels) and sand dune – floodplain boundaries. Coolabahs could grow in a wide range of soil types but favoured silt-sand soil types and rarely grew in floodplains dominated by cracking clays with well-developed gilgai structures. Relatively few recruits were identified from the sequence of large flood years in 2009-2011. There was evidence of considerable grazing pressure on the younger coolabahs and also plant moisture stress from the subsequent dry years of 2013-2015. The subdued recruitment response to the 2009-2011 flood sequence highlights the vulnerability of dryland riparian vegetation to grazing and climate change effects. 1
INTRODUCTION
Riparian and floodplain tree communities are often the dominant woodland assemblage in the arid zone. These communities have high ecological value in providing habitat, food sources and shade, and are a significant influence on the ecology of arid regions [1]. Coolabah (Eucalyptus coolabah) is an iconic Australian tree species and the dominant riparian species in many arid and semi-arid zone river systems, such as in the Lake Eyre Basin (LEB). The successful germination and growth (i.e. recruitment) of these trees, which occurs in response to floods [2, 3], is critical to the ecology of these dryland rivers. Understanding the drivers of recruitment success is vital to determining how these systems respond to current and future pressures driven by grazing, flow regulation and climate change. Previous work has examined regeneration characteristics of E. coolabah on the regulated rivers of the Darling River catchment, where clearing of this species has commonly occurred [3]. The unregulated rivers and the largely natural vegetation communities of the LEB provide an excellent system for analysis of natural patterns of tree distribution. This study examines field data on tree distribution, flooding characteristics and soil properties to identify drivers of successful germination and growth (i.e. recruitment) of E. coolabah in a large unregulated arid zone river, the Diamantina River of the LEB. The Diamantina River is one of the major rivers of the LEB and is the major contributor to flow into Lake Eyre North, the terminus of the endorheic LEB. It has a summer-dominant, intermittent flow regime and shows a variety of channel-floodplain morphologies, ranging from meandering single channel to vast areas of poorly channelized floodplain dominated by cracking clay sediments [4]. 2
METHODS
2.1 Field data collection Fieldwork was undertaken during a field trip in May 2014. Data on the distribution of E. coolabah trees were collected from 19 belt transects on the Diamantina floodplain. Transects typically began at the main channel and were directed approximately perpendicular to the channel direction and were of varying length (200-1200 m). All E. coolabah trees located within the 10 m width of the transect were measured for their location and elevation (by differential GPS), number of stems, circumference of the largest stem above ground level,
maximum height of tree canopy, maximum width of tree canopy and reproductive status (i.e. fruiting or not fruiting). In addition, between 2-6 soil profiles where drilled using a hand auger on each transect, totaling 54 soils profiles. Profiles were between 0.3-0.5 m deep with soil samples collected every 0.1 m. The soil profile locations were chosen to characterize soil conditions where E. coolabah trees were relatively abundant and where they were absent (i.e. no E. coolabah within 30 m radius of the soil profile). The Geoscience Australia Water Observations from Space (WOfS) dataset provided a measure of the frequency of inundation over the study area using the Landsat record. This dataset has a 25 m spatial resolution and was used to assign mean frequency of inundation values to individual tree locations in each transect (see Figure 1). Laboratory analysis was subsequently completed on the soil samples and included: gravimetric water content of a 100 g sample dried at 100°C for 24 hours; electro-conductivity (EC) and pH measurements taken from a 1:5 mix of dried soil sample (10 g) and deionized water (50 ml); plasticity measurements using a „rolltest‟ method. A small amount of the soil was mixed with a few drops of water until soil moisture content reached field texture (i.e. just below sticky point). The resulting mix was rolled into a cylinder by hand to a length of at least 40 mm. The cylinder was then held by the end to determine if it can support its own weight. The degree of plasticity categorises the soil samples, where a diameter of 2 mm gives a very plastic soil (scored 5) and a diameter of 6 mm is non-plastic (scored 1). Figure 1. Location of transects along the Diamantina River overlaying the Water Observations from Space (WOfS) mapping of frequency of inundation mapping.
Table 1. K-means cluster analysis results showing five clusters and mean characteristics for live E. coolabah individuals within transects. Cluster Circum. (m) Crown (m) Height (m) n
A
B
C
D
E
0.08 0.80 1.19 533
0.23 2.17 3.15 188
0.52 4.43 5.78 159
1.02 8.45 6.70 87
1.38 13.33 9.09 51
2.2 Data analysis Three different methods were used to analyse the datasets; principal component analysis (PCA), K-means cluster analysis and the Kolmogorov-Smirnov (KS) test. The PCA analysis was used to minimize the effects of covariance in identifying variables correlated with the tree distribution, the K-means cluster analysis identified natural groupings within the tree dataset, while the KS test analyses for differences between two populations without assumptions of normal distribution. A PCA was initially performed on each of the three dependent tree measurement variables (diameter, crown and height) and 16 independent variables (soil characteristics including salinity, pH, plasticity, soil water content; inundation frequency; elevation above local channel thalweg; distance to other trees). The variables were range-standardised to remove noise in the data and remove any scaling influences. Outliers were identified and removed from the dataset following an initial PCA analysis. The results of the covariance matrix were used to identify poorly correlated variables; which were also subsequently removed from the data in an attempt to increase the variability captured by the first three principal components. A PCA was then re-performed on the reduced dataset. Two sample, non-parametric Kolmogorov-Smirnoff (KS) tests were performed to further analyse the soil data. Two populations of soil data were tested for significance between variables; „soil samples with trees‟ and „soil samples without trees‟.
3
RESULTS AND DISCUSSION
The May 2014 field trip measured 1057 individual trees with circumferences ranging from 0.01-3.30 m, heights from 0.10-14.00 m and crown diameters from 0.02-24.30 m. Seedlings related to recent floods in the 2009-2012 period were not obvious to identify as the majority of individuals with short heights (e.g.
